/srv/irclogs.ubuntu.com/2014/09/04/#juju.txt

=== scuttle|afk is now known as scuttlemonkey
=== CyberJacob is now known as CyberJacob|Away
=== menn0 is now known as menn0-afk
=== menn0-afk is now known as menn0
* arosales finishing up review time03:51
fuzzyHi there I just build a juju deployment using the quick start and manual provisioning.  I'm using 12.04 and have the precise juju-gui installed, but when I do a juju status it says pending.  The machine is at 100% idle and it's been like 5 minutes.  Is there anyway to figure out what is holding it up?04:15
=== uru-bot is now known as urulama
=== CyberJacob|Away is now known as CyberJacob
=== mfa298___ is now known as mfa298
=== CyberJacob is now known as CyberJacob|Away
=== kaydo___ is now known as kaydo
stubInteresting... my agent died while in debug-hooks09:43
stublxc09:43
jam2stub: seems surprising, anything interesting in the log file?09:43
stubjam2: yeah, I seem to have a big arsed go panic in there09:45
stubjam2: here we go. http://pastebin.ubuntu.com/8231693/09:47
=== allenap_ is now known as allenap
=== dosaboy_ is now known as dosaboy
=== scuttle|afk is now known as scuttlemonkey
=== wwitzel3_ is now known as wwitzel3
aisraelWhich project contains the text of this page? Doesn't look like it's juju-docs. https://juju.ubuntu.com/13:30
tvansteenburghhow does one test an upgrade-charm hook?13:31
tvansteenburghnevermind, `juju upgrade-charm -h` look helpful13:33
lazyPowerhey  marcoceppi, hazmat, jcastro, mbruzek did you see we had a community member working with linode contribue a quick start stack script for creating a juju jump-host on linode? https://www.linode.com/stackscripts/view/1019313:36
hazmatlazyPower, i did13:36
jcastrothat is awesome13:37
jcastroI'll send him a shirt13:37
hazmatlazyPower, ideally it would be wrapped up into a plugin like the digitalocean provider using the api.. its pretty manual to use something like that13:37
hazmatbasically its restricted to a one node env without some work13:37
lazyPowerhazmat: well, I was working with fuzzy to achieve their goal of having a jump station for their trusted dev-staff to control deployments.13:38
hazmatnot sure if it really needs the host name set, or if could bootstrap localhost and be auto for the one node case13:38
jcastrooh you're talking to them already?13:38
lazyPowerthey didn't like the idea of porting ~/.juju - so they threw it up as a stack script, and thats effectively their manual jump host.13:38
lazyPowerjcastro: yeah - i've known fuzzy a little over a year. Fuz is working with our Meteor.js charm13:39
jcastrooh ok13:39
jcastro# Stephanie Sunshine <sms@ziphub.com>13:39
lazyPowerit started as "how can i gate deployments a-la capistrano style?" - to which I showed them Juju-strano13:39
lazyPowerThats the one.13:39
jcastrowhat about her? should I send her a  shirt or are there multiple ones?13:39
lazyPowerStephanie Sunshine = Fuz13:39
jcastroack13:40
jcastroI'll send along a link to the DO provider as well13:40
lazyPowerhazmat: I pinged the list with that stack script, any feedback for them on that thread would be awesome. I'll point fuz at it when she shows up later today.13:41
lazyPowerI was talking about a phaux provider like your DO plugin - but for PoC work that was overkill. I can see that being on the roadmap as a community contribution though - baby steps :)13:41
jcastroyeah13:42
aisraeljcastro: any idea where that page lives, project wise (see previous question)?13:44
jcastrothe website, it's a wordpress setup13:45
jcastrothere should be a l ink on where to report bugs at the bottom13:45
jcastro... or not13:45
jcastrohttps://launchpad.net/juju-website13:45
aisraelack, thanks13:46
aisraeljcastro: I assume content changes just need to have a bug filed against it?13:50
jcastroyeah13:51
jcastroif you want irc ping luca13:51
jcastrohe's just taken over the site and is doing a massive clean up/reorg13:51
aisraelkk13:54
natefinchmarcoceppi, jcastro, anyone else:  do I have to do open_port 80 in my install hook if I declare my app to be http, or will juju expose do that for me?14:23
marcoceppinatefinch: all port openings need to be explicit14:23
natefinchmarcoceppi: ok, that makes sense14:23
marcoceppiport 80 isn't the only port an http server could run on14:23
* marcoceppi wanders off pondering tomcat14:23
=== urulama is now known as urulama_afk
jcastronatefinch, the other half of that bug, which is annoying14:41
jcastrois the whole directory structure for local charms14:41
jcastroso like, juju expects:14:41
jcastro<trusty>14:42
jcastro-> <charmname>14:42
jcastroand if you do `juju deploy --repository=foo local:blah`14:42
jcastroif foo doesn't have a series/charm structure, it just bails14:42
natefinchjcastro: you should read my latest email about writing a charm, I mentioned that specifically14:43
jcastroyeah14:43
natefinchjcastro: my suggestion is to make juju deploy --local=<path>  work if <path> points to a charm directory14:43
jcastroyeah!14:43
jcastronatefinch, is there a specific bug for that or is it just on your radar somewhere?14:43
natefinchjcastro: on my radar, but we should make a bug14:43
natefinchjcastro: my radar is notoriously unreliable ;)14:44
jcastrook I'll copy and paste your text and toss it in there14:44
natefinchjcastro: awesome, thanks14:44
jcastrohttps://bugs.launchpad.net/juju-core/+bug/136554214:46
mupBug #1365542: Doesn't deploy a charm unless the repository directory has a specific directory structure <ui> <juju-core:New> <https://launchpad.net/bugs/1365542>14:46
natefinchjcastro: great14:49
=== fgallina is now known as fgallina-lunch
=== jog_ is now known as jog
=== katco` is now known as katco
=== fgallina-lunch is now known as fgallina
JoshStroblHey arosales_ where do you think the Code Review section in the Charms Review Doc should be? Prior to the charmers team part so it doesn't break apart the absolutely essential parts of the review?15:38
JoshStroblAlso, I was thinking about moving up the community review ending part to prior to the charmers team, that way the process of the charmers team can take up more room if necessary and not push down the community content.15:40
arosales_JoshStrobl: hmm, it would live in the workflow your defined, perhaps before deploy/test.15:41
arosales_JoshStrobl: or we could leave that specifically for the ~charmer section15:41
arosales_your thoughts?15:41
JoshStroblhmm15:42
arosales_I guess it comes back to if we think everyone will want/be able to do a code review.15:42
JoshStroblexactly15:42
JoshStroblI don't want to make it seem like it is only meant for the charmers team.15:43
arosales_ya +1 to that15:43
JoshStroblbut it is your call15:43
arosales_I think this should be _the_ review process15:43
arosales_we could put it out to the list for input too15:43
arosales_ideally, the review process is transparent and anyone wanting to can do a full reivew.15:44
arosales_but15:44
arosales_perhaps folks just want a workflow that is straight forward and does not include code review15:44
JoshStroblmarcoceppi, do you know if the "expanded sections" header thing (the Juju specific Markdown) works?15:45
JoshStroblarosales_, if the expanded sections part works, we could have it prior to the deployment section.15:45
JoshStroblarosales_, it could be unexpanded by default, marked optional, those that want or are required to dive into the code review can expand the section15:46
JoshStroblmarcoceppi, what I am referring to -> https://github.com/juju/docs/wiki/Markdown#foldouts15:47
marcoceppiJoshStrobl: yes, those all work15:48
marcoceppiJoshStrobl: you can see it in action on this page:15:48
marcoceppihttps://juju.ubuntu.com/docs/reference-release-notes.html15:48
JoshStroblah sweet, thanks15:48
JoshStroblarosales_, would that work?15:49
marcoceppiJoshStrobl: if you need additional HTML elements that don't exist yet, Im happy to make more markdown plugins15:49
allomovHey, all.15:50
allomovI will have short question about deploying to openstack.15:50
marcoceppiallomov: feel free to ask your question15:50
allomovI get following error trying to bootstrap openstack environment: https://gist.githubusercontent.com/allomov/188131c84fdbcbab5ea8/raw/b7192994bbab0f0c70d8e1a19f7dcb09cab3cb70/fail-juju.log15:50
allomovhere is a piece from my environment.yml file https://gist.githubusercontent.com/allomov/188131c84fdbcbab5ea8/raw/799c20a50083e82bf832cadb3e72ad7d2159c5a4/juju-env.yml15:51
marcoceppiallomov: you need to create cloud image metadata for your openstack provider15:51
marcoceppiallomov: https://juju.ubuntu.com/docs/howto-privatecloud.html15:51
arosales_JoshStrobl: I am thinking if we just put out code review section in the ~charmer section but welcome anyone to also do that as part of their reivew . ..15:52
allomovmarcoceppi: cool, thank you for the source15:52
arosales_JoshStrobl: but also try to keep this readable so folks don't think this is just for ~charmers15:52
marcoceppiallomov: the first half of that doc is pretty heavy with details that might confuse at first. Make sure you read through once :)15:52
allomovmarcoceppi: that's the strange thing, because I didn't need to use my region name before. fog gem for instance doesn't need it15:53
marcoceppiallomov: Juju does a bit more and things slightly differently than fog gem15:53
arosales_JoshStrobl: my 2 cents, fwiw, lets put the code review section in the "general/community" review process. Cavet the code review is optional pending review comfort level.15:55
arosales_inlucde some common things to look for that we can grow out so folks can grow their skill there if they want15:55
* arosales_ thinking out loud15:55
arosales_JoshStrobl: as you mentioned this could also be folded in too15:55
JoshStroblarosales_, sounds good to me.15:55
JoshStroblarosales_, I'll make note of the existing code review contents you pointed out in the comments and it can be expanded up later if others come up with more stuff, sound good?15:56
=== fabrice is now known as fabrice|dinner
arosales_JoshStrobl: sounds good16:01
JoshStroblFantastic. I'll get started on it now then!16:02
* arosales_ really liking this doc.16:02
arosales_I will also be creating a complimentary charm author guidliens doc specifically to call out design consideration and common gotchas in reviews that we can take care of before review.16:03
arosales_JoshStrobl: be good to get your input there if you have time given your a charm author16:03
JoshStroblarosales_, sounds good. do you want us to figure it out after your guys' sprint?16:04
arosales_JoshStrobl: ya I have this slated to creat at the sprint next week.16:08
JoshStrobl\o/16:08
=== hatch__ is now known as hatch
arosales_JoshStrobl: to conrim I read that correctly, re "figure it out after your guy's sprint" was that in regards to your charm review doc, or my charm author guidlines doc.16:12
JoshStroblarosales_, your charm author guidelines doc16:12
JoshStroblarosales_, we understand each other correctly?16:13
arosales_JoshStrobl: we do :-)16:16
JoshStroblgreat :)16:16
JoshStroblarosales_, sorry to nag you again, was thinking about another thing to watch out for in the code review, you approve of the following? "Not using set -ex in bash scripts, which allows the bash script to more easily catch failed execution, thereby allowing Juju to more accurately detect a failed hook."16:44
arosales_JoshStrobl: not nagging at all16:47
marcoceppiJoshStrobl: yes, that should def be in there16:52
marcoceppiJoshStrobl: well, only -e is needed16:52
marcoceppi-x is xtrace16:52
JoshStroblmarcoceppi, ah16:52
JoshStroblI'll remove the x part then :P16:53
marcoceppi-eux should be in best practices though, (-u being that if a varaible is accessed before set it's an error)16:53
JoshStroblmarcoceppi, so you think I should have it as -eux or something like "Not using set -e or set -eux[...]"16:54
marcoceppi-e is the minimum for a review, -eux should be in the best practice section of the docs imo16:54
* marcoceppi may make a merge for that16:54
=== uru_ is now known as urulama_afk
arosales_JoshStrobl: +1 to marcoceppi comments on just making sure -e is there16:59
arosales_+1 on recommending -eux though per marcoceppi's suggestions17:00
JoshStroblmarcoceppi, request for the script that converts the GFM / Juju Markdown into HTML (you know, the make stuff), have a sha1sum of each file. Do an on-the-fly comparison between the sha1sum of the files in src/en/ and the existing sha1sums. If the file changed, THEN convert it. Otherwise don't bother. I think it'd speed up the converting time when using "make" since it'd only convert files that have changed.17:02
JoshStrobl(does this with some of his custom "compilers")17:03
marcoceppiJoshStrobl: sure, that's a great idea. Can you open an issue on GH17:03
marcoceppiI'm collecting a list of things I can do "offline" the next time I'm flying17:03
JoshStroblyep, is the converter in juju/docs?17:03
marcoceppiJoshStrobl: yup17:04
JoshStroblcool, I'll file an issue there then17:04
marcoceppiyou're talking about a hashsum of the compiled file, right?17:04
marcoceppirather17:04
JoshStroblno, a sha1sum of the src/en markdown files17:04
marcoceppiif compiled file exists, and hashsum of the source hasn't changed, then skip17:04
marcoceppiotherwise, you'd never get an initial copy of the compiled file17:05
JoshStroblif src/en/*.md sha1sum !== existing sha1sum of the file (stored in a .sha1sum file), compile, else skip17:05
JoshStroblright, if the sha1sum simply doesn't, go ahead and compile it anyways, since the file probably doesnt exist in htmldocs17:06
marcoceppiJoshStrobl: yeah, i get teh concept. def open an issue, it's a great idea esp as the docs increase size17:06
marcoceppiI was going to implement a file watcher17:06
marcoceppiso as src/* files changed it would just recompile them17:06
JoshStroblthat'd be great too, but I'd prefer to not have to compile every single markdown file into HTML whenever I make a change I want to test :P17:07
JoshStroblbut yea, I'll file an issue in a bit17:07
marcoceppiit would only compile the markdown file that changed, not all the files17:08
marcoceppibut yeah17:08
JoshStroblYou can sorta get an idea of how I do it by looking at the recursiveCompiling function at https://github.com/StroblIndustries/Metis/blob/master/build/compiler/metisCompiler and the fileChanged function at https://github.com/StroblIndustries/Metis/blob/master/build/compiler/compilerFunctions17:10
JoshStroblit is bash, I'm going to assume yours is in python, but the logic is essentially the same17:11
avoineanyone got that error on a bootstrap?17:13
avoineERROR juju.cmd supercommand.go:323 cannot initiate replica set: cannot get replica set configuration: cannot get replset config: not authorized for query on local.system.replset17:13
JoshStroblarosales_, doc is now updated: https://github.com/JoshStrobl/docs/commit/bf62c234572b94ff61cec6709db3190654639433?diff=split17:13
marcoceppiavoine: what version of juju are you using? `juju version`?17:14
avoine1.20.6-trusty-amd64 and juju-mongodb 2.4.9-0ubuntu317:14
avoinefrom http://ppa.launchpad.net/juju/stable/ubuntu17:15
avoineand with an lxc local bootstrap17:16
* arosales_ loving the split in github17:17
arosales_JoshStrobl: thanks taking a look17:17
JoshStroblmarcoceppi, issue filed - https://github.com/juju/docs/issues/15717:21
marcoceppikirkland: using "run-one-constantly" in the new review queue, love these little helper scripts17:49
kirklandmarcoceppi: woot :-)17:49
kirklandmarcoceppi: btw, I just found, per dpb2, the timeout command, which is nice in conjunction with run-one*17:50
marcoceppinice17:50
dpb2:)17:50
=== CyberJacob|Away is now known as CyberJacob
=== roadmr is now known as roadmr_afk
=== CyberJacob is now known as CyberJacob|Away
arosales_marcoceppi: got a question for you18:55
=== arosales_ is now known as arosales
arosalesany any yaml experts out there18:55
arosalesI am looking at the cassandra charm18:55
marcoceppiarosales: what's the problem?18:55
arosalesand it is currently failing proff as it is missing defatult value for some of the keys18:56
arosalesI am looking for a recomendaton on what to put for a default18:56
arosalesis it simply, "default:"18:56
marcoceppiarosales: just put the default key, but leave it blank18:56
arosalesor18:56
marcoceppiyes, that one18:56
arosalesdefault: ""18:56
arosalesmarcoceppi: ack on just the default key, but blank, ie "default:"18:56
arosalesmarcoceppi: thanks.18:57
marcoceppito keep current behavior, you should use the former, if blank string is needed (intead of unset) then the latter18:57
arosalesmarcoceppi: ack.18:57
marcoceppiit's the difference between "" and NULL which most charms wont' care about, but some do18:57
arosalesI'll check that doesn't cause any issues on the charm deploy18:57
arosalesya I was wondering if some charms check for any value to be set18:57
arosalesand take action on that even if it is blank18:58
arosalesor specifically look for a Null, and then key off that18:58
arosalesthat was my main concern18:58
=== roadmr_afk is now known as roadmr
lazyPowerhatch: no. The mismatch in versions between deployed and what you have installed locally wouldn't cause that. (moving convo here)20:27
hatchahh ok, just noticed it, thought I'd point it out20:28
hatch:)20:28
hatchI'm really hoping you can reproduce this20:28
hatchheh20:28
lazyPowerhatch: i found the issue though20:28
lazyPoweryour source option is coming back as undefined. and it should be defaulting to true.20:28
hatchso is this a GUI bug?20:29
lazyPoweri'm pointing a finger at the devel branch of the gui,  I'm testing with -stable now and pending a deployment.20:29
lazyPowerit appears so20:29
hatchinteresting....20:30
hatchwow if it is, it's good we caught it now and not next week lol20:30
hatchlazyPower:  I'll also try on stable20:32
lazyPowertvansteenburgh: http://paste.ubuntu.com/8253401/20:54
lazyPoweri ran into a promulgate bug - any idea what i've snagged?20:54
lazyPowerwait, i found it. when i cloned it it was missing the push branch it was sniffing for. I overrode the lp resource to use the ~charmers route and it worked.20:55
=== CyberJacob|Away is now known as CyberJacob
wesleymasonSo...I have a theory that in a certain situation the containerName passed to OpenstackStorage.Remove() can be blank (perhaps if a destroy-environment is called after a failed destroy due to an API timeout, not sure), which will result in a DELETE being sent to the swift account URL (and in v1.7.x of swift marks the account on swift as deleted, which is a total PITA to then fix as it involves either a) waiting for a cron'd script to really delete the DB, or21:00
wesleymasonb) manually removing the sqlite3 DB file for the user on the swift replicas - both of which then results in the account being recreated based on the keystone perns for the user on next access. Now I have an easy fix for that (check for the error condition specifically in func Remove(...)), but not a clue how to replicate the issue, other than the fact that this is the only explanation I have for how my swift account got screwed after a failed destroy-enviro21:00
wesleymasonnment followed by a successful one.21:00
wesleymasonAny thoughts?21:00
wesleymasonprobably better off asked in #juju-dev, but more after ideas or reactions like "that sounds ridiculs, naaah"21:01
wesleymason*ridiculous21:02
hatchlazyPower: so I was able to deploy it just fine in an old version of the GUI21:17
hatchyou the same?21:17
hatchsource isn't even set when running the juju-run command21:18
hatcher, it's not in the list21:18
dpb2If I give an http:// url to image-metadata-url, it still downloads with 'https://' is that expected?21:24
dpb2and by downloads, I mean, downloads the images21:30
lazyPowerdpb2:are  you referring to simple streams?21:33
lazyPowerhatch: correct, it worked fine using a -stable revision of the gui.21:33
lazyPowerer, whatever is provided by default.21:34
lazyPowerhatch: i'd re-target the bug against the GUI and point to this LP resource for all the aggregated details.21:34
dpb2lazyPower: maybe, I'm referring to the images for lxc creation that get downloaded form cloud-images.ubuntu.com21:34
lazyPowerdpb2: ah, yeah it defaults to secure comms to avoid MITM attacks.21:34
dpb2lazyPower: but I changed it to http?21:35
dpb2no good?21:35
lazyPowernatefinch: are there any schenanigans with fetching of the cloud image url structure?21:35
lazyPowerdpb2: so, offtopic - i've noticed you've incremented by 1, level up?21:36
dpb2lazyPower: perhaps... you never know with me21:36
hatchlazyPower: definitely21:36
hatchthanks so much for the help21:36
dpb2lazyPower: I think freenode didn't like my identity. :)21:36
lazyPowerhatch: np :) I smelled some schenanigans once I saw the deploy configuration in the bug.21:37
lazyPowerdpb2:  Freenode, like camelot - is a silly place.21:37
=== dpb2 is now known as dpb1
hatchlazyPower: I've also got help on getting apache2 going for ghost and which options I need so very close to finally piecing together a bundle :)21:39
lazyPowerhatch: awesome! I look forward to seeing your bundle submission21:40
dpb1lazyPower: I'm back!21:41
lazyPowerdpb1: decrement21:41
=== pjdc_ is now known as pjdc
mbruzekarosales, Github will not let me merge your change as-is.22:27
JoshStroblmbruzek, what is the issue?22:27
arosalesmbruzek:ok let me get this conflict resolved22:27
arosaleshttps://github.com/juju/docs/pull/154 is in question22:28
mbruzekI have no problem with the *change* I need you to run the commands to get your branch current (is my guess).22:28
mbruzekJoshStrobl, I see the following message:   We can’t automatically merge this pull request.22:28
mbruzekto resolve conflicts before continuing.22:28
JoshStroblarosales, maybe your master branch of your fork isn't updated?22:29
* JoshStrobl checks22:29
JoshStroblyea it is 186 commits behind juju master22:29
fuzzyHi there, I got a new Juju installation going, but on the command and control node, my logs are being spammed pretty bad.  Could someone tell me what is going on and possibly how to fix it? https://clbin.com/XZ63i22:30
arosalesya it appears so22:30
mbruzekhi fuzzy I have not seen that before.22:30
* arosales working to update22:30
mbruzekfuzzy, what version of Juju are you using?22:31
arosalesmbruzek: I think if I push to master it will hit master22:34
arosalesmbruzek: to confirm your review was a +1?22:34
mbruzekarosales, Confirmed.  It looks great22:34
arosalesmbruzek: ok thanks22:35
fuzzySorry my wifi card likes to forget it exists22:35
mbruzekfuzzy, have not seen that error before what version of Juju are you using?22:35
fuzzythe latest one for 12.04 as of last night22:35
fuzzyI followed the quickstart for manual provisioning22:36
fuzzyis there an easy way to get juju to spit out version info?22:37
mbruzekfuzzy, juju version22:38
fuzzy1.20.7-precise-amd6422:38
mbruzekhrmm... That *is* the latest.22:39
mbruzekWhat command are you running that it gives you this log output?22:39
fuzzycat /var/log/syslog22:39
fuzzywell tail -n 100 /var/log/syslog22:39
* JoshStrobl is going reset his fork of juju/docs. git sometimes doesn't like to fast-forward so I magically end up with extra commits that are then ahead of juju/docs.22:40
mbruzekfuzzy, I have several messages like that in my syslog22:40
mbruzekfuzzy, so I guess that is normal.22:41
mbruzekfuzzy, You can go to #juju-dev and ask the developers about those logs, but I see them all over my syslog too.22:41
sarnoldwow that's noisy..22:42
mbruzekfuzzy, I am about to end my day I have to get to dinner.22:44
=== CyberJacob is now known as CyberJacob|Away
=== CyberJacob|Away is now known as CyberJacob
lazyPowerfuzzy: that looks like the migrations weren't run against the collection setting roles22:59
lazyPowerand yes, thats very noisy.22:59
fuzzyty :)23:00
fuzzyabout 10mb in 24 hours of logs23:00
=== CyberJacob is now known as CyberJacob|Away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!